AI Coding Agent Guardrails Secure the use of AI coding agents across engineering teams. This skill covers permission boundaries, secret protection, sandbox isolation, code review gates, and audit trails for Claude Code, Cursor, Copilot, and Codex. --- When to Use Apply these guardrails when: - Onboarding AI coding agents into an engineering team for the first time - Developers are using Claude Code, Cursor, Copilot, or Codex to generate production code - Agents have access to repositories containing secrets, infrastructure configs, or sensitive business logic - Your compliance framework (SOC…